Scoring Shots by Winning and Losing Teams

This visualisation shows the number of Scoring Shots (ie Goals + Behinds) registered by winning and losing teams in every game during a season, also with a single point for every combination in that season and with a 2d kernel density plot overlaid. (Note that, in drawn games, the Home team is treated as being the winning team for the purpose of this chart.)

Here's the same data, cumulated across seasons in the same was as the Scoring data earlier. Again, those gaps ... 

Excess Scoring Shots Registered by Winners

This visualisation shows, as a density plot for each season, the difference between the number of Scoring.Shots registered by the winning team and the number registered by the losing team. Overlaid on the plots are lines at the mean and at the point where the excess Scoring Shot figure is 0 (ie the cutoff point, below which the winning team registered fewer Scoring Shots than the losing team) and where it is 15. Summary statistics are provided in the upper right corner. (Note that, in drawn games, the Home team is treated as being the winning team for the purpose of this chart.)

Own and Opponent Scoring Shot Production, and Own Winning Rate in Past 50 Games

Now we look at the Scoring Shots - goals plus behinds - scored and conceded by teams in their previous 50 games, Home and Away or Finals, here for the period from 2000 to 2024.

A team's winning rates over those same 50 games is encoded in the colour of its label. The strong relationship between generating scoring shots, not conceding them, and winning is apparent, as we tend to find teams whose labels are red in the bottom right corner, and teams whose labels are blue in the top left corner.

... and now we do it for the entire history of the sport at the time of posting (ie from when a team first has 50 games behind it to 2024).

Relationship Between Conversion Rate and Scoring Shots!

By Era

We know that winning teams tend to have higher conversion rates (goals as a proportion of scoring shots) than losing teams, but do teams that register more scoring shots also tend to register a greater percentage of them as goals? Does practice make perfect?

There is only mild evidence that teams generating more Scoring Shots tend to be more accurate, but the effect size is quite small

Note that the spread of Conversion Rates reduces with the number of Scoring Shots (ie if we think about conversion as sampling from a Bernoulli distribution, we're seeing the effects of increasing the sample size, n, and so reducing the variance).

By Team (2000-2024)

We next turn out attention to the possibilty that, while there is no strong relationship when we aggregate across teams, perhaps we’ll find something at the team level for the most recent era.

The evidence is here too very weak for any meaningful (linear) relationship between scoring shots and conversion rates, here looking at the team level. Only Adelaide (2.8%), Essendon (1.7%), and St Kilda (1.2%) have r-squared of more than 1%.

By Venue (2000-2024)

Lastly, let’s investigate the relationship between scoring shots and conversion rates for venues with at least 25 games worth of data in the modern era (so, data for at least 50 teams).

Here, too, there is little evidence for any strong (linear relationships) with only Bellerive Oval, Football Park, Marrara Oval, Princes Park, and Sydney Showground showing an r-squared above 1%.